Biotic invasion has become a critical issue in most ecosystems word-wide. Non-native species invasions can lead to numerous adverse consequences for native systems. Invasion can modify ecological properties, processes, and functions [1]. Invasive plants can compete with native species for resources, disrupt evolutionary processes and hybridize with natives, reduce system productivity, alter disturbance regimes, and threaten native biodiversity [1,2,3,4,5]. Additionally, invasive species can impose large economic costs, through loss of economic output and the cost of control [6,7,8]. The threat of invasion continues to increase due to globalization, world transport and travel, and climate change [9,10,11,12].

Invasion in southeastern U.S. forestlands poses a severe threat to an ecologically and economically important region. Southeastern forests contain over 200 million acres of timberland, providing 62% of U.S. timber production and over 15% of world-wide timber [13,14]. Invasive plants threaten the continued productivity of important hardwood and softwood species which is essential to the economic sustainability of the region [15]. In addition, Southeastern U.S. forests contain a high level of native biodiversity, with over 3000 plant species [16,17,18], and provide many other important ecosystem services.

Japanese honeysuckle (Lonicera japonica Thunb.), a perennial woody vine, is one of the more ubiquitous and aggressively invasive pests of southeastern U.S. forests. Although introduced in the early 1800s, L. japonica was first recorded outside of cultivation along the Potomac River in 1882 [19]. Numerous different cultivars have subsequently been introduced and it has presently become naturalized in 45 states of the U.S. and the Canadian province of Ontario [20,21]. L. japonica has many characteristics which make it a successful invader. It experiences rapid growth [22], phenotypic plasticity for response to herbivore damage, leaf phenology, and plant growth form and structure [23,24], evergreen or semi-evergreen leaves, and early flowering for long durations [22]. Additionally, it can reproduce vegetatively or through seed, and is accessible to a wide diversity of pollinators (both diurnal and nocturnal due to dusk flower opening). It exploits light gaps efficiently [23] and can resprout from the roots following disturbance [25,26].

L. japonica is found in most southeastern forest communities and can have negative effects on the native vegetation of the communities in which it is prevalent [27]. It has been shown to decrease diversity [28,29,30], modify vegetation structure [31], and inhibit growth and reproduction of native species under certain circumstances [32,33,34]. There are currently no predators or pathogens in the introduced range which have significant detrimental effects on L. japonica [20], and eradication from an area requires constant intensive combinations of removal strategies [20,31].

Although already wide-spread, L. japonica continues to expand its range in North America and invade formerly intact areas within its established range [20]. The frequency of these occurrences is likely to increase with increased temperatures [35,36,37] and additional disturbances [38,39]. Reliable predictions of habitats at risk are needed in order to enact preventative measures for controlling the spread of L. japonica. In this paper, we apply logistic regression to analyze an extensive data set collected as part of the Forest Inventory and Analysis (FIA) program of the U.S. Forest Service. We aim to identify potential determinants of invasion and predict the likelihood of further invasion under a variety of possible management strategies.

Environmental niche models which include current distribution and land use parameters can be valuable tools for predicting the future spread of L. japonica. Although L. japonica is naturalized and found in most habitat types in southeastern U.S. forests [20], it is still spreading into uninvaded areas possessing suitable environmental conditions [62,63]. Predicting the likelihood of invasion into a given area will allow managers to focus prevention measures on high risk and environmentally sensitive areas. Additionally, correlation determinations between occurrence and management can assist managers in choosing appropriate strategies for minimizing risk in areas environmentally suited to further L. japonica invasion. Our modeling analysis suggests that while L. japonica distribution is ultimately driven by a suite of environmental characteristics, certain management strategies might be effective for preventing its spread.

Suitable environmental characteristics are the dominant correlates with L. japonica occurrence, but anthropogenic features and land use/management variables also showed strong negative and positive correlations with potential invasion. Proximity to water bodies, daily maximum temperatures, and site productivity were all strongly positively correlated environmental variables. Climate is one of the primary determinants of species range [64] and invasion is frequently found to be highest in climatically suitable areas [65]. However, climate alone is not sufficient to determine a species’ invasion potential [66]. Stand age, artificial forest regeneration, and fire disturbance are negatively correlated, whereas private land ownership is positively correlated with probability of invasion, suggesting that land management activities contribute substantially to successful invasions.

Artificial regeneration, while expensive, can decrease the risk of L. japonica invasion. This might be an important consideration in areas at high risk, given the potential economic losses attendant to L. japonica invasion. L. japonica, like other invasive vines, can inhibit growth of host species through light interception and even kill host species through girdling [67]. L. japonica has been shown to inhibit reproduction and growth of non-host trees through root competition around forest edges [34]. In addition, L. japonica can indirectly inhibit growth of native species through site occupation and seedling suppression [30,33,68]. These negative effects manifest themselves as lost profits in forests managed for timber. Additionally, extensive and oftentimes cost prohibitive measures are required to eradicate L. japonica once established. With artificial regeneration following timber harvest, stands are intensively managed for desirable species, which maintains conditions that are more appropriate for the native desired species and less appropriate for L. japonica [69]. This usually includes site preparation through root-raking and burning or disking [70] which reduces likelihood of L. japonica establishment into the newly created forest gap. This removal of non-desirable species and subsequent rapid stand growth can inhibit the establishment of L. japonica into a newly formed forest gap, which is otherwise highly at risk for invasion given the aggressive exploitation of light gaps and rapid growth under high light conditions exhibited by L. japonica [23,32]. Intense disturbance can create favorable habitat for L. japonica [71]. However, planting native species and subsequent management for those species reduces the opportunity for L. japonica spread. Also, the resulting single-aged stand has lower potential for future small gap dynamics which L. japonica might exploit, given that most of the trees will survive to harvestable age and L. japonica invasion is severely restricted in closed canopy conditions [72].

While fire alone has not been an effective restoration technique for the eradication of L. japonica from an invaded area, our model showed that the probability of invasion is negatively correlated with fire disturbance, suggesting that prescribed burning might be an important tool in reducing the spread of L. japonica into uninvaded forests. Fire alone can reduce above-ground L. japonica density temporarily, but it readily resprouts from the roots allowing it to rapidly replace damaged tissues following a typical prescribed fire [26,28,73], although repeated fires and high intensity fire have been shown to reduce overall L. japonica abundance in several studies [74,75,76]. Management of uninvaded areas with prescribed fires can decrease the likelihood of invasion. Fire eliminates vegetative spread from invaded areas to surrounding forests. Additionally, prescribed burning can cause mortality in newly established seedlings of resprouting woody plants because the young plants have not had time to establish carbon reserves sufficient to allow root sprouting and the loss of photosynthetic tissues can lead to subsequent carbon starvation if reserves are low enough at the time of burning [77,78]. This is corroborated by West et al. [24] who determined that L. japonica can be suppressed by fire after observing higher L. japonica abundance in areas of high litter cover associated with escape from fire and Brewer and Menzel [76], who found that a combination of thinning and frequent burning reduces L. japonica abundance.

Several land-use features are also correlative indicators of potential distribution. We included land-use features in our model because current and past land-use has been shown to affect the distribution and abundance of many invasive plants [79,80]. Additionally, Surrette and Brewer determined that land use changes can actually create habitat suitable for L. japonica which disfavors native vegetation, leading to an increase in the abundance of the invader relative to native species [71]. Distance to the nearest road is negatively correlated with occurrence. Roads can act as invasion corridors facilitating the spread of invasive plants. This is especially likely for an invasive vine such as L. japonica which rapidly grows vegetatively in light gaps (like those created along roadways). This growth is facilitated by the dual growth forms characteristic of many vine species [19]. Where there is no existing structure for twining, L. japonica can grow in a mat form along the ground, increasing the likelihood that roads could act as corridors for, rather than barriers to, spread. Additionally, L. japonica is an important forage species for wildlife and was cultivated for planting [81]. Most plantings likely occur close to human populations which are usually located near roads, so increased invasion along roadways was likely facilitated by cultivation in areas near roads and along roads themselves. This is corroborated by other studies which show increased L. japonica invasion in areas of higher human population [21]. Indeed, Pysek et al. [82] present evidence suggesting a strong relationship between human distributions and invasive plants generally. We also found a highly positive correlation between private land ownership and invasion potential. This is likely the result of more intensive and expensive forest management occurring on public land, but could also be a result of more recent cultivation on private lands for wildlife forage, and potential size differences in forest parcels owned publically and privately. The smaller parcels owned by private landowners are more susceptible to invasion due to the fragmentation issues noted above. Baars et al. [83] found a higher threat of invasion in smaller forest remnants with smaller interior areas than in larger tracts with less edge relative to interior forest. While converting areas to public ownership is not a feasible management option for the prevention of L. japonica invasion, understanding the differences in management that lead to the higher invasion probabilities for publically owned lands can assist in the development of management recommendations for private landowners, which, if followed, could help avoid further spread of L. japonica on private lands without a switch in ownership.

In addition to looking at current climatic and environmental correlates of invasion potential and exploring management effects on risk of invasion, we investigated the potential for climate change to alter the invasive potential of L. japonica. In general, climate change and its attendant habitat alterations can facilitate spread and increased abundance of invasive species [84,85] and lead to changes in spatial patterns of invasive distribution [86]. While temperature effects on L. japonica have not been experimentally explored, an increase in overall global temperatures would yield an expected increase in the area of climatically suitable habitat for L. japonica. In addition, our model showed an increase in overall risk of invasion, and an increase in the area of highest risk, with a 2 °C increase in annual temperature, resulting from the strong positive correlation between annual maximum temperature and invasion potential. This suggests that not only will the potential range of L. japonica increase with climate change, but the likelihood of invasion increases within the existing range distribution. This information is important for forest managers in areas with currently low invasion probabilities that might increase with climate change in the future.

While niche models such as ours are useful for determining invasion potentials for large areas, there are some shortcomings of this modeling approach. First, our model is based on current presence and absence data, which might not represent the entire potential niche of the species, since invasion is an ongoing process [66,87] and adaptation of L. japonica to its new environment may expand its potential niche [88]. Second, we modeled several potential drivers of invasion only implicitly. We included biotic interactions implicitly via inclusion of diversity and productivity correlates of L. japonica occurrence, but particular species interactions likely help shape the invasion process [89].

Nonetheless, we feel our model offers important insights into the management of L. japonica, which threatens to continue its invasion throughout southeastern U.S. forests, as well as other, recently invaded, areas [90]. Armed with knowledge of current and potential future high risk areas and, hence, likely paths of invasion, forest managers can develop long term monitoring and control strategies for effectively slowing range expansion and mitigating its effects.